We have been working on optimizing further the application. There is a new Linux application out which is able to actively select the device architecture.
Speed for compute capability 1.2,1.3 should be higher now. At least two digits percent improvements.

Windows application is under internal testing.

GDF.

It still takes the whole CPU core by default. After manual "renice" it looks as it should in my opinion. Will see how about speed in this version.

---

From Siberia with love!

I've had one:
No CPU time improvement but 12% faster on my GPU (GTX275) than the 6.04.
Not bad !!!

Very good.
gdf

I've got a 6.15 beta in the cache with about 3.5 hours to go on a production WU. Is it preferable to suspend the production WU and run the beta immediately or take it in order?

Beta app 6.15 WU f113-TONI_KK20BETA-5-100-RND3209_1 successfully completed but only about 3% faster than an app 6.04 equivalent. CPU usage same as app 6.04. Some improvement anyway. I have another beta waiting to start in about 3 hours.

Platform: Phenom II x4, 4GB, Fedora 11, GTS-250 1GB, 6.10.17

GTS-250 is based on the G92 GPU core, so it does not gain anything from the new application. Only G200 cores run faster.

gdf

I've had one:
No CPU time improvement but 12% faster on my GPU (GTX275) than the 6.04.
Not bad !!!

I had another one.
I confirm the speed improvement. (10-12% on my GTX275)

GTX275, ubuntu 9.10 x64. CPU usage is the same (100%), although no doubt it's faster than 6.04

Guys, could you pls look closer on CPU usage?

---

It will be very useful even if someone could just provide a way to make GPU apps run at lowest priority (so as CPU apps) without necessarity to modify it's priority manually...

---

From Siberia with love!

me personally put GPUGRID at reasonably high priotiry (I wanna get more points, ha-ha). But 100% CPU usage killing my CPU-based POEM project...

---

That's exactly what I am talking about...

---

From Siberia with love!

6.15 A69-TONI_KKBETA2-0-100-RND8994_0 task errored out after 10 hours
ID 64724 - 9600GT, Ubuntu 64bit, 100% load of one core

<core_client_version>6.10.30</core_client_version>
<![CDATA[
<message>
process exited with code 193 (0xc1, -63)
</message>
<stderr_txt>
# There is 1 device supporting CUDA
MDIO ERROR: cannot open file "restart.coor"
# There is 1 device supporting CUDA
# There is 1 device supporting CUDA
SWAN : FATAL : Failure executing kernel sync [PmeRealSpace_compute_forces_kernel] [702]
acemdbeta_6.15_x86_64-pc-linux-gnu__cuda: swanlib_nv.cpp:221: void swanRunKernel(const char*, block_config_t, block_config_t, size_t, ...): Assertion `0' failed.
SIGABRT: abort called
Stack trace (14 frames):
../../projects/www.gpugrid.net/acemdbeta_6.15_x86_64-pc-linux-gnu__cuda(boinc_catch_signal+0x4d)[0x46265d]
/lib/libc.so.6[0x7f3a67e2a040]
/lib/libc.so.6(gsignal+0x35)[0x7f3a67e29fb5]
/lib/libc.so.6(abort+0x183)[0x7f3a67e2bbc3]
/lib/libc.so.6(__assert_fail+0xe9)[0x7f3a67e22f09]
../../projects/www.gpugrid.net/acemdbeta_6.15_x86_64-pc-linux-gnu__cuda[0x46ff13]
../../projects/www.gpugrid.net/acemdbeta_6.15_x86_64-pc-linux-gnu__cuda[0x45b6eb]
../../projects/www.gpugrid.net/acemdbeta_6.15_x86_64-pc-linux-gnu__cuda[0x458a54]
../../projects/www.gpugrid.net/acemdbeta_6.15_x86_64-pc-linux-gnu__cuda[0x4294b0]
../../projects/www.gpugrid.net/acemdbeta_6.15_x86_64-pc-linux-gnu__cuda[0x41292c]
../../projects/www.gpugrid.net/acemdbeta_6.15_x86_64-pc-linux-gnu__cuda(sin+0xbf2)[0x407ffa]
../../projects/www.gpugrid.net/acemdbeta_6.15_x86_64-pc-linux-gnu__cuda(sin+0x3b4)[0x4077bc]
/lib/libc.so.6(__libc_start_main+0xe6)[0x7f3a67e155a6]
../../projects/www.gpugrid.net/acemdbeta_6.15_x86_64-pc-linux-gnu__cuda(sinh+0x51)[0x407519]

Exiting...

</stderr_txt>
]]>